top of page

Yesterdays Card >

Model Architecture

Our six-factor scoring system is the foundation of our predictive edge. Each module is weighted to reflect its impact on the final outcome.

Starting Pitching

30%

The primary determinant of a game. We analyze velocity, movement, and historical strikeout rates to project the starter's performance.

Bullpen

20%

Impact of the relievers. We evaluate the bullpen's depth and the probability of a blown save, factoring in the opposing team's offensive strength.

Offense vs Starter

25%

The matchup between the opposing team's hitters and the starting pitcher's zone. We calculate the expected number of outs and runs.

Park/Weather

10%

Environmental factors. We adjust for ballpark dimensions, wind speed, and temperature to project how the game will play out.

Situational

10%

Specialized play. We account for specific game situations like the bases loaded or a runner on second, adjusting for high-leverage scenarios.

Injury/Lineup

5%

Player availability and lineup adjustments. We factor in injury reports and bench depth to project the most likely starting XI.

Edge Thresholds & Sizing

Technical documentation on the model's decision logic for edge thresholds and confidence tier sizing.

Edge Threshold
Confidence Tier
Model Logic

3.0% Edge

1.5u

High-probability edge with significant value potential.

1.5% Edge

1.0u

Medium-probability edge with moderate value potential.

0.5% Edge

0.75u

Low-probability edge with minimal value potential.

0.1% Edge

0.5u

High-risk, low-reward edge with significant volatility.

bottom of page